Table 2 Classification accuracy after training with different datasets.

From: Environmental sound classification using temporal-frequency attention based convolutional neural network

Class

No mask

Mask I

Mask A

AC

91.5%

90.0%

77.2%

CH

90.9%

88.3%

82.8%

CP

86.3%

82.1%

75.5%

DB

88.0%

85.2%

76.6%

DR

90.1%

82.9%

72.8%

EI

92.8%

93.9%

76.9%

GS

94.7%

91.2%

84.9%

JA

91.0%

85.6%

80.6%

SI

93.3%

90.7%

79.4%

SM

89.1%

84.7%

76.6%

Ave

90.8%

87.4%

78.3%

Back to article page